Accession Number : ADA134187
Title : Bit Weighting and Soft Decision Demodulation for Image Compression & Error Control.
Descriptive Note : Final rept. May 79-Dec 82,
Corporate Author : TEXAS A AND M UNIV COLLEGE STATION DEPT OF ELECTRICAL ENGINEERING
Personal Author(s) : Gibson,J D ; Griswold,N C
PDF Url : ADA134187
Report Date : Aug 1983
Pagination or Media Count : 209
Abstract : There are numrous operational requirements for remotely sensed imagery. The vast amounts of image data acquired in these applications pose a particular problem for transmission to other locations. In order to achieve real-time operation and to conserve communications channel bandwidth, it is common to compress the image data. Suitable image compression techniques reduce the required transmitted data rate by a factor of approximately 8-to-1, while still maintaining acceptable image quality. Unfortunately, when an image is compressed, it is much more susceptible to bit errors introduced by noisy channels. This report document research concerning image compression at 1 bit/pel over noisy channels. The techniques of bit weighting, soft decision demodulation, and channel coding are compared for reducing or eliminating the effects of channel errors on compressed imagery. The source coding method used is the two-dimensional discrete cosine transform of 16 by 16 blocks. Results indicate that for an independent bit error rate of 0.01, bit weighting alone and soft decision alone provide a noticeable improvement in reconstructed image quality. Combined bit weighting/soft decision outperforms either technique alone. Judicious channel coding can substantially improve image quality over noisy channels with only a slight degradation in ideal channel performance.
Descriptors : *Data compression, *Image processing, *Error correction codes, Acoustic channels, Noise reduction, Gaussian noise, Monte Carlo method, Two dimensional, Coefficients, Methodology, Comparison, Optimization, Algorithms, Data transmission systems
Subject Categories : Cybernetics
Distribution Statement : APPROVED FOR PUBLIC RELEASE